[7918] in Kerberos

home help back first fref pref prev next nref lref last post

Re: security on compromised networks (Fwd)

daemon@ATHENA.MIT.EDU (Dan Geer)
Thu Sep 5 11:29:14 1996

To: rdf@swan.admin.ccny.cuny.edu, giff@traveller.east.sun.com
Cc: members@sug.org, kerberos@MIT.EDU
In-Reply-To: Your message of "Thu, 05 Sep 1996 10:38:49 EDT."
             <Roam 0.9.2.841934329.27453.giff@traveller> 
Date: Thu, 05 Sep 1996 11:09:35 -0400
From: Dan Geer <geer@OpenMarket.com>


Wayne/Robert,

Transport level security is a fine idea when an application
is hopeless, you have the juice to crypt the entire data stream,
and you're willing to trust conventional usernames and passwords
for authentication.  But I must add that Kerberos-modified versions
of telnet, rsh, rlogin, POP, NFS and many, many more are widely
available, some as freeware and some as commercial products.  For
that reason, I have included the Kerberos mailing list on my reply.

--dan

reference message for others new to this conversation:

    Robert,
    
    Your message was forwarded to me.

    > We have been having a series of breakins lately and it is
    > obvious that the person has been sniffing ethernet packets
    > to obtain passwords.
    > 
    > It seems to me that this is going to have to be an standard
    > assumption on all networks in the future since people with
    > PC will be able to examine packets and the concept of a root
    > user with a small, know group of trusted people will no longer
    > be true.
    > 
    > The weakest links that I know of are telnet, su during a telnet
    > session and using POP for accessing email. All these protocols
    > send passwords in clear text. Even if one is careful never to
    > use telnet for sensitive accounts and never to su to root 
    > from a remote machine, end user's accounts will be insecure and
    > subject to unauthorized access.
    > 
    > Using an add-on security mechanism such as Kerberos does not
    > seem to be practical since it is impossible to dictate the sort
    > of client software that people will use.
    > 
    > Does anyone have suggestions about these issues? Thoughts,
    > experience appreciated. Thanks...
    
    My group at Sun (the Internet Commerce Group) has just recently 
    begun shipping a product targeted exactly at your stated problem.
    
    The product is SKIP for Solaris. It provides an implementation of
    the SKIP protocol which provides transparent IP layer encryption
    between SKIP compliant hosts.
    
    We are currently shipping only for Solaris (Sparc and X86) and should have
    Windows based versions soon. 
    
    SKIP is a key management protocol for IP and provides multiple options
    for packet level encryption from 40 bit RC-4 to Triple DES.
    
    For more information or to arrange for an evaluation,
    please contact me at giff@incog.com or call me at
    703-240-4929
    
    You can also visit our Web site at http://skip.incog.com
    and even download the beta version of Solaris SKIP for
    eval purposes.
    
    giff


home help back first fref pref prev next nref lref last post